INTRODUCTIONMutation of the transcription factor GATA3 in humans causeshypoparathyroidism, sensorineural deafness and renal dysplasia(HDR) syndrome (Bilous et al., 1992; Van Esch et al., 2000), whichdisplays a high degree of phenotypic heterogeneity. Manyindividuals with the mutation do not display the full HDR triadand across patients the severity of defects varies widely and caninclude palatal and central nervous system defects (Barakat et al.,1977; Bilous et al., 1992; Ferraris et al., 2009; Fujimoto et al., 1999;Fukami et al., 2011; Hasegawa et al., 1997; Lichtner et al., 2000;Muroya et al., 2001; Van Esch et al., 2000). A variety of GATA3mutations have been described in humans, with differing effectson the function of GATA3 (Nesbit et al., 2004). However, there areno clear genotype-phenotype correlations for HDR (Adachi et al.,2006; Zahirieh et al., 2005). Rather, there is a substantial amountof intrafamilial variation, which has been suggested to possibly bedue to genetic background effects (Fukami et al., 2011; Hernándezet al., 2007; Mino et al., 2005; Nakamura et al., 2011; Zahirieh etal., 2005). However, the cause of this variation is still unknown.

The incomplete penetrance and highly variable expressivity ofHDR syndrome suggest a level of canalization: that, in manyinstances, development is robust enough to overcome reductions

in GATA3 levels. HSP90 activity associates with canalization ofphenotypes and disease resistance across diverse taxa (Aridon etal., 2011; Chen and Wagner, 2012; Gangaraju et al., 2011; Lu et al.,2003; Queitsch et al., 2002; Yeyati et al., 2007). Large bodies ofevidence show that HSP90 is involved in numerous cellularactivities, including protein folding (Johnson, 2012; Taipale et al.,2010). Because there are several missense mutations that causeHDR syndrome, HSP90 activity is a candidate for regulating someof the variability observed in HDR.

Models to investigate the variability of HDR are lacking, althoughmouse models for all aspects of HDR have been generated (Duncanet al., 2011; Grigorieva et al., 2010; Haugas et al., 2012; Karis et al.,2001; Lilleväli et al., 2006; Lim et al., 2000; van der Wees et al.,2004). Here, we describe a zebrafish point mutation in gata3 thatis homologous to a mutated site in human HDR (Nesbit et al., 2004).We show that gata3 mutant zebrafish display the HDR triad andhave craniofacial defects, the severity of which vary significantlydepending upon genetic background. Furthermore, we providenovel insights into the interplay between gata3, Ahsa1, Hsp90 andthe generation of variability in zebrafish gata3 mutant phenotypes.

RESULTSIn a forward genetic screen for zebrafish craniofacial mutants, weisolated the b1075 mutant allele. Using PCR-based genetic mappingof linkage to simple sequence length polymorphisms (SSLPs), wefound tight linkage to z20450 on linkage group 4, with no crossoversout of 196 meioses, and placed the mutation in an interval betweenz6977 and z11657, with 2 and 11 crossovers, respectively. Finermapping positioned b1075 in an ~325 kb interval between 1075-11 and 1075-8, each with one crossover in 552 meioses. Thisinterval contained five predicted genes: itih2, similar to kin, atp5c1,taf3 and gata3. Sequence analysis uncovered a thymidine toadenosine point mutation within exon 4 of gata3 (supplementary

Disease Models & Mechanisms 1285

Disease Models & Mechanisms 6, 1285-1291 (2013) doi:10.1242/dmm.011965

1Department of Molecular and Cell and Developmental Biology, Institute forCellular and Molecular Biology, Patterson 522, University of Texas at Austin, Austin,TX 78713, USA2School of Life Science, Peking University, 5 Summer Palace Road, Beijing 100871,China*These authors contributed equally to this work‡Author for correspondence (eberhart@austin.utexas.edu)

Received 31 January 2013; Accepted 27 May 2013

© 2013. Published by The Company of Biologists LtdThis is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distributionand reproduction in any medium provided that the original work is properly attributed.

SUMMARY

The severity of most human birth defects is highly variable. Our ability to diagnose, treat and prevent defects relies on our understanding of thisvariability. Mutation of the transcription factor GATA3 in humans causes the highly variable hypoparathyroidism, sensorineural deafness and renaldysplasia (HDR) syndrome. Although named for a triad of defects, individuals with HDR can also exhibit craniofacial defects. Through a forwardgenetic screen for craniofacial mutants, we isolated a zebrafish mutant in which the first cysteine of the second zinc finger of Gata3 is mutated.Because mutation of the homologous cysteine causes HDR in humans, these zebrafish mutants could be a quick and effective animal model forunderstanding the role of gata3 in the HDR disease spectrum. We demonstrate that, unexpectedly, the chaperone proteins Ahsa1 and Hsp90 promotesevere craniofacial phenotypes in our zebrafish model of HDR syndrome. The strengths of the zebrafish system, including rapid development, genetictractability and live imaging, make this an important model for variability.

Ahsa1 and Hsp90 activity confers more severecraniofacial phenotypes in a zebrafish model ofhypoparathyroidism, sensorineural deafness and renal dysplasia (HDR)Kelly Sheehan-Rooney1,*, Mary E. Swartz1,*, Feng Zhao2,*, Dong Liu2 and Johann K. Eberhart1,‡

RESEARCH REPORTD

iseas

e M

odel

s & M

echa

nism

s

DM

M

material Fig.  S1A), resulting in a predicted cysteine to serinemissense mutation in the zinc ion (Zn2+)-coordinating domain ofzinc finger 2 (supplementary material Fig. S1B,C). Injection of agata3 morpholino phenocopied the b1075 mutant (supplementarymaterial Fig.  S1D-F), validating that b1075 is a mutant allele ofgata3. The cysteine that is mutated in b1075 is homologous to acysteine that is mutated in some cases of human HDR (Nesbit etal., 2004), suggesting that zebrafish could be a HDR model.

We characterized gata3 expression to determine whether thetissues disrupted in HDR expressed gata3 in zebrafish. Neural crestcells within the fate map region that is destined to become thetrabeculae (Eberhart et al., 2006; Swartz et al., 2011) expressed gata3(supplementary material Fig.  S1G, arrow). In amniotes, thepharyngeal endoderm of the second through to the fourth archgenerates the parathyroids (Grigorieva et al., 2010; Okabe andGraham, 2004). Although zebrafish embryos lack parathyroids, by72 hpf the pharyngeal endoderm expressed gata3 in zebrafish(supplementary material Fig. S1H). This expression extended fromthe third arch posteriorly, consistent with the zebrafish expressionof gene homologs involved in parathyroid specification in amniotes(Hogan et al., 2004; Okabe and Graham, 2004). The ear and the

embryonic kidney (pronephros) both expressed gata3 at 33 hpf(supplementary material Fig. S1I,J). The corpuscles of Stannius alsoexpressed gata3 and are functionally related to the parathyroids inregulating Ca2+ (supplementary material Fig.  S1J). These resultsshow that the evolutionary homologs and functional analogs of thetissues affected in human HDR express gata3 in zebrafish.

To test for functional conservation of GATA3 between zebrafishand human, we examined HDR tissue homologs in our zebrafishmutant. Our analyses were performed across two different geneticbackgrounds, WIK and EkkWill, and, although phenotypes acrosswild-type embryos did not vary (and are combined for statistics),substantial variation was observed in mutants, across backgrounds(Fig. 1). We refer to the WIK and EkkWill backgrounds as the ‘mild’and ‘severe’ backgrounds, respectively, because WIK mutants wereconsistently milder. We initiated our analyses on the craniofacialskeleton because this is where we found the greatest degree ofvariability.

The neurocranium of the zebrafish (Fig. 1A-C) lies immediatelyventral to the brain. The anterior neurocranium, or the zebrafishpalatal skeleton, is positioned medial to the eyes and consists of amidline ethmoid plate and bilateral trabeculae. In the mildbackground, the trabeculae were consistently present, whereas, ina severe background, the trabeculae were consistently absent(Fig.  1B,C). Quantification of the number of intacttrabeculae/embryo demonstrated that wild type, mild mutants andsevere mutants averaged 2 (n=654), 1.7 (n=125) and 0.17 (n=58)trabeculae, respectively (Fig.  1P). The difference in the averagenumber of trabeculae/embryo between mild and severe mutantswas extraordinarily significant as determined via Student’s t-test(P=4×10−40). Interestingly, the variation around the mean wassimilar within both mild and severe mutants, with standarddeviation (s.d.)=0.55 and 0.46, respectively. In wild-type embryos,chondrocytes within the trabeculae were stacked one upon anotherin a largely single-file fashion (Fig.  1D, asterisks). Although thetrabeculae were present, chondrocytes fail to stack appropriatelyin mild mutants (Fig. 1E, asterisks). Additionally, in 100% of mildand severe mutants, the lateral commissure connected to thetrabeculae, instead of its normal more posterior position (Fig. 1A,B,arrowheads).

We next analyzed gill buds because they are evolutionarily relatedto and require the gene regulatory networks involved in parathyroiddevelopment in amniotes (Hogan et al., 2004; Zajac and Danks,2008). Gill bud length was decreased in both mild and severe gata3mutants (Fig. 1G-I, outlined). Whereas wild-type embryos averaged12.2 gill buds (s.d.=0.75, n=20), mild and severe mutants averaged7.9 (s.d.=2.35, n=12) and 4.8 (s.d.=2.12, n=14), respectively. Thisdifference in gill bud number between mutants was significant(P=0.0018).

We stained sensory hair cells within the ear via anti-MyoVIantibodies (Fig. 1J-L) to test for ear defects. The average numberof sensory hair cells in wild-type embryos was 10.2 (s.d.=1.70,n=12). Although mutants had a reduced number of sensory haircells relative to wild type, mild and severe mutants did notsignificantly vary relative to each other, averaging 7.5 (s.d.=0.85,n=10) and 6.88 (s.d.=1.25, n=8) cells, respectively. We labeled thepronephros with ret1 (Fig. 1M-O) to test for renal defects. ret1 wasexpressed in the pronephros of mild mutants but absent in severemutants (Fig.  1N,O). Collectively, these data show that, like in

dmm.biologists.org1286

Craniofacial phenotypes in a zebrafish model of HDRRESEARCH REPORT

TRANSLATIONAL IMPACT

Clinical issuePhenotypic variability is a common feature of congenital malformations (birthdefects). Although the genetic underpinnings of a large number of birthdefects are beginning to be understood, the mechanisms underlying thisclinical variability remain unclear. HDR (hypoparathyroidism, sensorineuraldeafness, renal dysplasia) syndrome, which is characterized by a variety ofcraniofacial defects, is an autosomal dominant condition caused by mutationsin the gene encoding a human zinc finger transcription factor, GATA3. Thesyndrome is difficult to treat because of the associated clinical diversity.Studies have revealed no correlation between GATA3 genotype and diseaseseverity, and there are currently no animal models available to investigate thecause of the inherent variability in symptoms.

ResultsUsing a forward genetic screen for craniofacial mutants, the authors identifieda gata3 mutant in zebrafish. The mutant has a missense mutation that disruptsthe first cysteine of the second zinc finger domain of Gata3. Mutation of thehomologous cysteine has been shown to cause HDR syndrome in humans. Theauthors report that zebrafish gata3 mutants have defects affecting the palatalskeleton, ear, embryonic kidney and the gill buds, which are evolutionarilyrelated to the human parathyroid. They show that the severity of thesephenotypes is highly dependent upon genetic background. Interestingly, theauthors reveal a role for chaperone proteins Ahsa1 and Hsp90 (heat shockprotein 90) in mediating the variability in craniofacial defects, at least partially.

Implications and future directionsThe work provides a novel, effective animal model for studying clinicalvariability in HDR and related syndromes. Zebrafish gata3 mutants survivethrough organogenesis, providing researchers who are interested in HDRsyndrome with a resource for analyzing the complete spectrum of defectswithin the same embryo. Importantly, the genetic-background-specificphenotypic differences provide a means for understanding the cause ofphenotypic variation. The genetic conservation between human and zebrafishcombined with the unique genetic tools available for zebrafish manipulationcould be leveraged to test the function of human GATA3 mutations in differentzebrafish backgrounds and to characterize the chaperone pathways that seemto regulate the severity of phenotypes caused by gata3 mutation in the modelsystem.

Dise

ase

Mod

els &

Mec

hani

sms

D

MM

human, mutation of gata3 causes variable defects, making zebrafisha tractable model to understand variability within HDR syndrome.To test our ability to modulate HDR phenotypes, we examined thefunction of pathways with known involvement in disease variabilityand canalization.

Across a wide range of taxa an important modulator ofphenotypic variability and canalization is heat shock protein 90(HSP90) (Aridon et al., 2011; Chen and Wagner, 2012; Gangarajuet al., 2011; Lu et al., 2003; Queitsch et al., 2002; Yeyati et al., 2007),making HSP90 a promising candidate to modulate HDR phenotypicvariability. We focused on the craniofacial phenotype because it isstrikingly canalized in the mild background. We treated zebrafishembryos with the HSP90 inhibitor 17AAG to determine whetherHsp90 was involved in the across-background variability.Surprisingly, we found that inhibition of Hsp90 caused the partialrestoration of the trabeculae in the severe genetic background(Fig. 2) but did not alter the craniofacial phenotype of mild mutants(data not shown). Hsp90 inhibition did not fully rescue themorphology of the anterior neurocranium, but significantly(P=0.001) increased the number of trabeculae that developed andfused to the posterior neurocranium in severe mutants (Fig. 2D).Treated mutants averaged 0.95 trabeculae (s.d.=0.94, n=20) and

untreated mutants averaged 0.12 trabeculae (s.d.=0.34, n=39). Thisresult demonstrates that Hsp90 activity promoted more severegata3 mutant phenotypes.

Although most evidence points to Hsp90 buffering againstsevere phenotypes, evidence in cystic fibrosis suggests that AHSA1-mediated HSP90 activation might correlate with a poorer prognosis(Wang et al., 2006). To determine whether Ahsa1 might beregulating gata3 mutant phenotypes in zebrafish, we performedqPCR to compare ahsa1 mRNA levels across genetic backgrounds.In an initial characterization, we found that ahsa1 was elevated3.4-fold in severe mutants relative to mild mutants (data not shown).To confirm and extend this finding, we tested ahsa1 levels inmutants and wild types across both backgrounds. Similarly, wefound that ahsa1 was elevated 3.2-fold in severe mutants relativeto mild mutants and that these differences seem to largely be dueto differences in expression levels across the genetic backgrounds(Fig. 3A).

To test the functional significance of this expression difference,we injected ahsa1 splice-site-blocking morpholinos (Fig.  3B) atdoses that left wild-type embryos unaffected (Fig. 3C). These levelsfailed to alter mild mutant phenotypes (Fig. 3D), but improved thephenotype of severe mutants (Fig. 3E,I). Uninjected severe mutants

Disease Models & Mechanisms 1287

Craniofacial phenotypes in a zebrafish model of HDR RESEARCH REPORT

Fig. 1. Genetic background influences the gata3 mutant phenotypes. (A-C)Flat-mounted neurocrania and (D-F) close-up views of the trabeculae (tr). In both(A) wild-type embryos and (B) mild gata3 mutants, the palate (p) is fully formed. However, there are rearrangements to the stacking of chondrocytes in mildmutants (asterisks D and E). (C,F)In severe mutants, the trabeculae are lost, generating a gap between the ethmoid plate (ep) and posterior neurocranium. (G-I)gata3 mutation disrupts outgrowth of gill buds (outlined). (H)Mild mutants generate fewer gill buds. (I)In severe mutants, the number of gill buds is furtherreduced. (J-L)Myosin-VI (MyoVI) labels sensory neurons (arrowheads) in the zebrafish ear. (K,L)Mild and severe mutants have fewer MyoVI -positive cells. (M-O)Wild-type and mutant embryos stained with the ret1 riboprobe. (M,N)ret1 expression is maintained in mild mutants (arrowheads). (O)In severe mutants,ret1 expression is absent. (P-R)Quantification of the defects in gata3 mutant embryos. All graphs show means ± 1 s.e.m. (P)The number of trabeculae perembryo are significantly reduced in severe mutants (average=0.17, s.e.m.=0.06, s.d.=0.46, n=58) compared with mild mutants (average=1.7, s.e.m.=0.05,s.d.=0.55, n=125). Wild-type embryos average=2 trabeculae (s.e.m.=0, s.d.=0, n=654). (Q)The number of gill buds per embryo is also significantly reduced insevere mutants (average=4.79, s.e.m.=0.57, s.d.=2.12, n=14) relative to mild mutants (average=7.92, s.e.m.=0.68, s.d.=2.35, n=12). Wild-type embryosaverage=12.15 gill buds (s.e.m.=0.17, s.d.=0.75, n=20). (R)Although there is a reduction compared with wild type (average=10.17, s.e.m.=0.49, s.d.=1.70, n=11),the number of MyoVI-positive cells are not significantly altered across mild (average=7.5, s.e.m.=0.27, s.d.=0.85, n=10) and severe (average=6.88, s.e.m.=0.44,s.d.=1.25, n=8) mutants. Anterior to the left; (A-F) dorsal views of flat-mounted neurocrania; (G-O) lateral views of whole-mounted embryos.

Dise

ase

Mod

els &

Mec

hani

sms

D

MM

averaged 0.45 trabeculae (s.d.=0.52, n=11), whereas ahsa1-morpholino-injected severe mutants averaged 1.09 trabeculae(s.d.=0.83, n=11, P=0.046). This result shows that Ahsa1 is necessaryto promote severe craniofacial phenotypes in gata3 mutants.

To test whether Ahsa1 was sufficient for severe phenotypes, weinjected mRNA encoding human AHSA1 at levels that did not affectwild-type embryos (Fig. 3F). We found that AHSA1 injection causedloss of the trabeculae in mild mutants (Fig. 3G). Uninjected mildmutants averaged 1.85 trabeculae (s.d.=0.35, n=25), whereasAHSA1-injected mild mutants averaged 0.62 trabeculae (s.d.=0.65,n=14; P=0.000013; Fig.  3G,J), suggesting that the Ahsa1-Hsp90pathway is more active in severe mutants.

Ahsa1-Hsp90 activity sequesters mutant CFTR, thus loweringits effective concentration (Wang et al., 2006). If the effectiveconcentration of Gata3 is different across genetic backgrounds thena graded morpholino injection should recapitulate phenotypesobserved in both backgrounds. Indeed, we found that inbred ABembryos injected with 5 ng of a gata3 morpholino very closelyresembled mild mutants, with alteration to the attachment of thelateral commissure (Fig. 4B, arrow) and cell stacking of thetrabeculae (Fig.  4B�, arrowheads) but with mostly attachedtrabeculae (Fig.  4B,B�,D). Embryos injected with 15 ng ofmorpholino had loss of the trabeculae that closely resembled severegata3 mutants (Fig. 4C-D). Collectively, our data demonstrate that,as in human, mutation of gata3 in zebrafish causes a highly variableset of defects. These data demonstrate that the AHSA1-HSP90pathway is involved in generating genetic-background-dependent

variation in zebrafish gata3 mutants and suggest that zebrafish willaid in understanding variability of human HDR.

DISCUSSIONMutation of GATA3 causes HDR syndrome in humans. We showthat tissues disrupted in HDR are defective in our zebrafishmutants. Our zebrafish mutation is in the homologous residue tohuman C318. In HDR this cysteine can be replaced by arginine(Nesbit et al., 2004), whereas in our zebrafish model a serine isgenerated. Although the serine replacement is more conservative,it is still predicted to result in a loss of Zn2+ coordination withinthe second zinc finger domain, which requires the cysteine. Indeed,

dmm.biologists.org1288

Craniofacial phenotypes in a zebrafish model of HDRRESEARCH REPORT

Fig. 2. Inhibition of Hsp90 partially restores trabeculae in severe mutants.(A)Zebrafish embryos were treated with levels of 17AAG that did not disruptthe craniofacial skeleton in wild type. (B)Untreated gata3 mutants in thesevere background typically lack trabeculae (arrowheads). (C)Hsp90 inhibitionpartially restored the trabeculae, resulting in the fusion of the palate andposterior neurocranium. (D)17AAG treatment of severe gata3 mutantssignificantly increases the number of trabeculae in treated embryos(average=0.95, s.e.m.=0.211, s.d.=0.94, n=20) compared with control, DMSO-treated, embryos (average=0.13, s.e.m.=0.05, s.d.=0.34, n=39). (A-C)Dorsalviews of flat-mounted neurocrania, anterior to the left; ep, ethmoid plate; tr,trabeculae.

Fig. 3. Ahsa1 is necessary and sufficient to promote severe gata3phenotypes. (A)ahsa1 expression is higher in gata3 mutants and wild-typeembryos in the severe background. (B)The ahsa1 morpholino causesmisplicing of ahsa1 mRNA [arrowhead in morpholino (MO) lane; 1 kb=1 kbplus ladder]. (C,D)In both wild-type and mild mutant embryos ahsa1

morpholino injection did not disrupt the trabeculae. (E)Knocking down ahsa1

in severe mutants resulted in a partial restoration of the trabeculae (arrow).(F)Wild-type embryos injected with low levels of AHSA1 mRNA display nocraniofacial defects. (G,H)Under these conditions, mutants from the (G) mildbackground closely resemble those from the (H) severe background, withtrabeculae loss (arrows). (I)Ahsa1 loss-of-function significantly increases thenumber of trabeculae in severe gata3 mutants (average=1.09, s.e.m.=0.25,s.d.=0.83, n=11), relative to uninjected control mutants (average=0.455,s.e.m.=0.16, s.d.=0.52, n=11). (J)Injection of AHSA1 mRNA significantly reducesthe number of trabeculae in mild gata3 mutants (average=0.615, s.e.m.=0.18,s.d.=0.37, n=13), relative to uninjected mutants (average=1.85, s.e.m.=0.1,s.d.=0.65, n=25). (C-H)Flat-mount images; anterior to the left.

Dise

ase

Mod

els &

Mec

hani

sms

D

MM

an in-depth analysis of human GATA3 mutations showed that anydisruptions to the second zinc finger domain resulted in a loss ofDNA binding (Nesbit et al., 2004). Future experiments will be aimedat determining whether the zebrafish mutation behaves similarlyin these types of analyses.

Our results demonstrate that activation of the Hsp90 pathwayis deleterious for zebrafish gata3 mutants. This was surprising giventhe clear role of Hsp90 in protecting against deleterious phenotypes(Gangaraju et al., 2011; Lu et al., 2003; Queitsch et al., 2002; Yeyatiet al., 2007). Although HSP90 seems to function predominantly bystabilizing proteins (Taipale et al., 2010), in cystic fibrosis Ahsa1downregulation enhances CFTR activity (Wang et al., 2006). Thedeleterious activity of HSP90 in cystic fibrosis is thought to be dueto HSP90 sequestration of a hypomorphic CFTR. Therefore, in ourzebrafish model Hsp90 sequestration might block functions thatthe missense Gata3 protein retains. Although the human GATA3C318R mutation fails to bind DNA in vitro (Nesbit et al., 2004), itis possible that, in vivo, higher-order protein complexes compensatefor this reduced DNA-binding capacity. Our future analyses willinclude testing the function and localization of human and mutantforms of GATA3 protein across genetic backgrounds in zebrafish.

The functional role of Gata3 that is buffered in the mild mutantsis still not understood. In severe gata3b1075 mutants, neural crestcells that should form the trabeculae become mislocalized (M.E.S.and J.K.E., unpublished). This, coupled with the stacking defect inmild mutants, suggests that Hsp90 activity modulates a role forGata3 in neural crest cell movements underlying palatogenesis.Analysis of neural crest cell movements across genetic backgroundswill provide important insights into the role of Gata3 in craniofacialdevelopment.

MATERIALS AND METHODSZebrafish care and husbandryZebrafish care protocols were IACUC approved and performedaccording to Westerfield (Westerfield, 1993). 17AAG stock wasdissolved in DMSO and applied as described (Yeyati et al., 2007).The gata3b1075 allele was generated through ENU-mediatedmutagenesis in a Tubigen background. The same female carrier wascrossed to the EkkWill and WIK genetic backgrounds. Geneticmapping was performed in the WIK background. The F1 offspringof these crosses exhibited dramatically different phenotypes (asdescribed) and, through continuous incrossing, these phenotypicdifferences have consistently been maintained.

Genotyping of gata3 was performed using primers: jke71 (f: 5�-GGAAACAGAAGGGGATGGGG-3�) and jke72 (r: 5�-TCTTACTAGAGAAGTGTAAGACAGCTAGGG-3�), followed byrestriction digestion with NlaIII: the mutant allele is 273 bp, andthe wild-type allele is 138 and 139 bp.

Staining protocols5-dpf zebrafish larvae were stained with Alcian Blue and AlizarinRed (Walker and Kimmel, 2007). For in situ hybridization andimmunohistochemistry, embryos were fixed in 4% PFA between32 and 72 hpf. The gata3 and ret1 riboprobes are described (Neaveet al., 1995; Wingert et al., 2007).

A rabbit polyclonal antibody was used at 1:500 to analyze MyoVIprotein expression (Proteus Biosciences, 25-6791). Embryos werefixed in 4% PFA overnight at 4°C. Embryos were washed with PBSthen water before treating with 100% acetone for 10  minutes at−20°C. Embryos were washed with water then PBS and thenincubated in blocking solution, containing 2% normal goat serumin PBDTx (PBS, 1% BSA, 1% DMSO, 0.5% Triton-X), for 1 hour atroom temperature. Anti-MyoVI antibody in blocking solution wasapplied overnight at 4°C. Following washing, a 1:200 dilution ofthe secondary antibody (Alexa-Fluor-568, Invitrogen) in PBDTxwas applied for 5 hours, at room temperature.

All embryos used for skeletal staining or in situ hybridizationwere imaged on a Zeiss Axioimager. Embryos analyzed byimmunohistochemistry were imaged on a Zeiss 710 confocalmicroscope.

Injections and pharmaceutical treatmentsInjections were performed at the one-cell stage. The E4I4 ahsa1morpholino (5�-TTAGAGCAGTCACCTGTTTTGAGAT-3�;Gene Tools) targets the splice site between exon4 and intron4. A3 nl bolus of an 8 mg/ml morpholino solution was injected. Theprimer pair ksr82 (5�-CCCAGCACAGCTAATGCTCC-3�) withksr83 (5�-TGCTGGCCAACTAGCAAACC-3�) assessedmorpholino efficacy. The translation blocking gata3 morpholino5�-CCGGACTTACTTCCATCGTTTATTT-3� (Gene Tools) (Yanget al., 2010) was used.

Tol2 competent human AHSA1 (Orfeome, Invitrogen) wascloned into pCSDest (Villefranc et al., 2007) following themanufacturer’s instructions (LR Clonase, Invitrogen).AHSA1:pCSDest was linearized with ApaI and mRNA wastranscribed using the Sp6  mMessage Machine kit (Ambion). A 3 nl injection of a 300 ng/μl stock of AHSA1 mRNA was injected.

Disease Models & Mechanisms 1289

Craniofacial phenotypes in a zebrafish model of HDR RESEARCH REPORT

Fig. 4. Graded injection of gata3 morpholino recapitulates thebackground-specific gata3 mutant phenotypes. (A,A�) Uninjected (UIC)embryos have normal neurocrania and trabeculae. (B,B�) Injection of 5 nggata3 morpholino results in the lateral commissure attaching to thetrabeculae (arrow, B) and improper stacking of cells within the trabeculae(arrowheads, B�). (C,C�) Injection of 15 ng of gata3 morpholino causestrabeculae loss. (D)Quantification of the average number of trabeculae/embryo.

Dise

ase

Mod

els &

Mec

hani

sms

D

MM

qPCR and statistical analysesStudent’s t-tests were performed between mild mutants and severemutants. All graphs show the mean ± 1 s.e.m. qPCR was performedin triplicate with Sybr Green (ABI) on a Viia7 system (Invitrogen)according to the manufacturers’ protocols. At 36 hpf, embryos weregenotyped using DNA extracted from tails, and heads were storedin RNAlater (Qiagen). Following genotyping, nine heads fromgata3−/− and gata3+/+ embryos from each background were pooledand RNA was extracted in Trizol reagent (Invitrogen) followed byDNase treatment and purification (RNAeasy MinElute Cleanup Kit,Qiagen). ahsa1 was amplified using primers ksr101 (5�-ACAGAGTTCGCTCAGGGTAT-3�) and ksr87 (5�-GCGCCCATCCACAAAAGCAGC-3�), and normalized to rpl13a(Tang et al., 2007).ACKNOWLEDGEMENTSThe authors thank Briana Schroeder, Jenna Rozacky and Melissa Griffin for fishcare.

COMPETING INTERESTSThe authors declare that they do not have any competing or financial interests.

AUTHOR CONTRIBUTIONSK.S.-R. and M.E.S. characterized the expression of gata3, variation across gata3

mutants, and performed the Hsp90 and Ahsa1 experiments. M.E.S. performed thegata3 morpholino experiments. D.L. and F.Z. provided the MyoVI staining protocoland the gata3 morpholino, and provided advice on using the morpholino. J.K.E.performed the genetic mapping, oversaw the conception and design of theexperiments, and wrote the paper.

FUNDINGThis work was supported by National Institutes of Health DE018088 and aUniversity of Texas Research Grant to J.K.E., and National Basic Research Programof the Chinese Ministry of Science and Technology 973 Grant 2012CB944503 toD.L. The gata3 mutant was obtained through a forward genetic screen supportedby National Institutes of Health HD22486.

SUPPLEMENTARY MATERIALSupplementary material for this article is available athttp://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.011965/-/DC1

REFERENCESAdachi, M., Tachibana, K., Asakura, Y. and Tsuchiya, T. (2006). A novel mutation in

the GATA3 gene in a family with HDR syndrome (Hypoparathyroidism, sensorineuralDeafness and Renal anomaly syndrome). J. Pediatr. Endocrinol. Metab. 19, 87-92.

Aridon, P., Geraci, F., Turturici, G., D’Amelio, M., Savettieri, G. and Sconzo, G.(2011). Protective role of heat shock proteins in Parkinson’s disease. Neurodegener.

Dis. 8, 155-168.Barakat, A. Y., D’Albora, J. B., Martin, M. M. and Jose, P. A. (1977). Familial nephrosis,

nerve deafness, and hypoparathyroidism. J. Pediatr. 91, 61-64.Bilous, R. W., Murty, G., Parkinson, D. B., Thakker, R. V., Coulthard, M. G., Burn, J.,

Mathias, D. and Kendall-Taylor, P. (1992). Brief report: autosomal dominant familialhypoparathyroidism, sensorineural deafness, and renal dysplasia. N. Engl. J. Med. 327,1069-1074.

Chen, B. and Wagner, A. (2012). Hsp90 is important for fecundity, longevity, andbuffering of cryptic deleterious variation in wild fly populations. BMC Evol. Biol. 12,25.

Duncan, J. S., Lim, K. C., Engel, J. D. and Fritzsch, B. (2011). Limited inner earmorphogenesis and neurosensory development are possible in the absence ofGATA3. Int. J. Dev. Biol. 55, 297-303.

Eberhart, J. K., Swartz, M. E., Crump, J. G. and Kimmel, C. B. (2006). Early Hedgehogsignaling from neural to oral epithelium organizes anterior craniofacialdevelopment. Development 133, 1069-1077.

Ferraris, S., Del Monaco, A. G., Garelli, E., Carando, A., De Vito, B., Pappi, P., Lala,R. and Ponzone, A. (2009). HDR syndrome: a novel “de novo” mutation in GATA3gene. Am. J. Med. Genet. A. 149A, 770-775.

Fujimoto, S., Yokochi, K., Morikawa, H., Nakano, M., Shibata, H., Togari, H. andWada, Y. (1999). Recurrent cerebral infarctions and del(10)(p14p15.1) de novo inHDR (hypoparathyroidism, sensorineural deafness, renal dysplasia) syndrome. Am. J.

Med. Genet. 86, 427-429.

Fukami, M., Muroya, K., Miyake, T., Iso, M., Kato, F., Yokoi, H., Suzuki, Y.,Tsubouchi, K., Nakagomi, Y., Kikuchi, N. et al. (2011). GATA3 abnormalities in sixpatients with HDR syndrome. Endocr. J. 58, 117-121.

Gangaraju, V. K., Yin, H., Weiner, M. M., Wang, J., Huang, X. A. and Lin, H. (2011).Drosophila Piwi functions in Hsp90-mediated suppression of phenotypic variation.Nat. Genet. 43, 153-158.

Grigorieva, I. V., Mirczuk, S., Gaynor, K. U., Nesbit, M. A., Grigorieva, E. F., Wei, Q.,Ali, A., Fairclough, R. J., Stacey, J. M., Stechman, M. J. et al. (2010). Gata3-deficient mice develop parathyroid abnormalities due to dysregulation of theparathyroid-specific transcription factor Gcm2. J. Clin. Invest. 120, 2144-2155.

Hasegawa, T., Hasegawa, Y., Aso, T., Koto, S., Nagai, T., Tsuchiya, Y., Kim, K. C.,Ohashi, H., Wakui, K. and Fukushima, Y. (1997). HDR syndrome(hypoparathyroidism, sensorineural deafness, renal dysplasia) associated withdel(10)(p13). Am. J. Med. Genet. 73, 416-418.

Haugas, M., Lilleväli, K. and Salminen, M. (2012). Defects in sensory organmorphogenesis and generation of cochlear hair cells in Gata3-deficient mouseembryos. Hear. Res. 283, 151-161.

Hernández, A. M., Villamar, M., Roselló, L., Moreno-Pelayo, M. A., Moreno, F. andDel Castillo, I. (2007). Novel mutation in the gene encoding the GATA3 transcriptionfactor in a Spanish familial case of hypoparathyroidism, deafness, and renal dysplasia(HDR) syndrome with female genital tract malformations. Am. J. Med. Genet. A. 143,757-762.

Hogan, B. M., Hunter, M. P., Oates, A. C., Crowhurst, M. O., Hall, N. E., Heath, J. K.,Prince, V. E. and Lieschke, G. J. (2004). Zebrafish gcm2 is required for gill filamentbudding from pharyngeal ectoderm. Dev. Biol. 276, 508-522.

Johnson, J. L. (2012). Evolution and function of diverse Hsp90 homologs andcochaperone proteins. Biochim. Biophys. Acta 1823, 607-613.

Karis, A., Pata, I., van Doorninck, J. H., Grosveld, F., de Zeeuw, C. I., de Caprona, D.and Fritzsch, B. (2001). Transcription factor GATA-3 alters pathway selection ofolivocochlear neurons and affects morphogenesis of the ear. J. Comp. Neurol. 429,615-630.

Lichtner, P., König, R., Hasegawa, T., Van Esch, H., Meitinger, T. and Schuffenhauer,S. (2000). An HDR (hypoparathyroidism, deafness, renal dysplasia) syndrome locusmaps distal to the DiGeorge syndrome region on 10p13/14. J. Med. Genet. 37, 33-37.

Lilleväli, K., Haugas, M., Matilainen, T., Pussinen, C., Karis, A. and Salminen, M.(2006). Gata3 is required for early morphogenesis and Fgf10 expression during oticdevelopment. Mech. Dev. 123, 415-429.

Lim, K. C., Lakshmanan, G., Crawford, S. E., Gu, Y., Grosveld, F. and Engel, J. D.(2000). Gata3 loss leads to embryonic lethality due to noradrenaline deficiency ofthe sympathetic nervous system. Nat. Genet. 25, 209-212.

Lu, R., Malcuit, I., Moffett, P., Ruiz, M. T., Peart, J., Wu, A. J., Rathjen, J. P.,Bendahmane, A., Day, L. and Baulcombe, D. C. (2003). High throughput virus-induced gene silencing implicates heat shock protein 90 in plant disease resistance.EMBO J. 22, 5690-5699.

Mino, Y., Kuwahara, T., Mannami, T., Shioji, K., Ono, K. and Iwai, N. (2005).Identification of a novel insertion mutation in GATA3 with HDR syndrome. Clin. Exp.

Nephrol. 9, 58-61.Muroya, K., Hasegawa, T., Ito, Y., Nagai, T., Isotani, H., Iwata, Y., Yamamoto, K.,

Fujimoto, S., Seishu, S., Fukushima, Y. et al. (2001). GATA3 abnormalities and thephenotypic spectrum of HDR syndrome. J. Med. Genet. 38, 374-380.

Nakamura, A., Fujiwara, F., Hasegawa, Y., Ishizu, K., Mabe, A., Nakagawa, H.,Nagasaki, K., Jo, W. and Tajima, T. (2011). Molecular analysis of the GATA3 gene infive Japanese patients with HDR syndrome. Endocr. J. 58, 123-130.

Neave, B., Rodaway, A., Wilson, S. W., Patient, R. and Holder, N. (1995). Expressionof zebrafish GATA 3 (gta3) during gastrulation and neurulation suggests a role in thespecification of cell fate. Mech. Dev. 51, 169-182.

Nesbit, M. A., Bowl, M. R., Harding, B., Ali, A., Ayala, A., Crowe, C., Dobbie, A.,Hampson, G., Holdaway, I., Levine, M. A. et al. (2004). Characterization of GATA3mutations in the hypoparathyroidism, deafness, and renal dysplasia (HDR)syndrome. J. Biol. Chem. 279, 22624-22634.

Okabe, M. and Graham, A. (2004). The origin of the parathyroid gland. Proc. Natl.

Acad. Sci. USA 101, 17716-17719.Queitsch, C., Sangster, T. A. and Lindquist, S. (2002). Hsp90 as a capacitor of

phenotypic variation. Nature 417, 618-624.Swartz, M. E., Sheehan-Rooney, K., Dixon, M. J. and Eberhart, J. K. (2011).

Examination of a palatogenic gene program in zebrafish. Dev. Dyn. 240, 2204-2220.Taipale, M., Jarosz, D. F. and Lindquist, S. (2010). HSP90 at the hub of protein

homeostasis: emerging mechanistic insights. Nat. Rev. Mol. Cell Biol. 11, 515-528.Tang, R., Dodd, A., Lai, D., McNabb, W. C. and Love, D. R. (2007). Validation of

zebrafish (Danio rerio) reference genes for quantitative real-time RT-PCRnormalization. Acta Biochim. Biophys. Sin. (Shanghai) 39, 384-390.

van der Wees, J., van Looij, M. A., de Ruiter, M. M., Elias, H., van der Burg, H.,Liem, S. S., Kurek, D., Engel, J. D., Karis, A., van Zanten, B. G. et al. (2004).

dmm.biologists.org1290

Craniofacial phenotypes in a zebrafish model of HDRRESEARCH REPORTD

iseas

e M

odel

s & M

echa

nism

s

DM

M

Hearing loss following Gata3 haploinsufficiency is caused by cochlear disorder.Neurobiol. Dis. 16, 169-178.

Van Esch, H., Groenen, P., Nesbit, M. A., Schuffenhauer, S., Lichtner, P.,Vanderlinden, G., Harding, B., Beetz, R., Bilous, R. W., Holdaway, I. et al. (2000).GATA3 haplo-insufficiency causes human HDR syndrome. Nature 406, 419-422.

Villefranc, J. A., Amigo, J. and Lawson, N. D. (2007). Gateway compatible vectors foranalysis of gene function in the zebrafish. Dev. Dyn. 236, 3077-3087.

Walker, M. B. and Kimmel, C. B. (2007). A two-color acid-free cartilage and bone stainfor zebrafish larvae. Biotech. Histochem. 82, 23-28.

Wang, X., Venable, J., LaPointe, P., Hutt, D. M., Koulov, A. V., Coppinger, J.,Gurkan, C., Kellner, W., Matteson, J., Plutner, H. et al. (2006). Hsp90 cochaperoneAha1 downregulation rescues misfolding of CFTR in cystic fibrosis. Cell 127, 803-815.

Westerfield, M. (1993). The Zebrafish Book: A Guide For The Laboratory Use Of Zebrafish(Brachydanio rerio). Eugene, OR: Institute of Neuroscience, University of Oregon.

Wingert, R. A., Selleck, R., Yu, J., Song, H. D., Chen, Z., Song, A., Zhou, Y., Thisse,B., Thisse, C., McMahon, A. P. et al. (2007). The cdx genes and retinoic acid control the positioning and segmentation of the zebrafish pronephros. PLoS Genet.3, e189.

Yang, L., Rastegar, S. and Strähle, U. (2010). Regulatory interactions specifyingKolmer-Agduhr interneurons. Development 137, 2713-2722.

Yeyati, P. L., Bancewicz, R. M., Maule, J. and van Heyningen, V. (2007). Hsp90selectively modulates phenotype in vertebrate development. PLoS Genet. 3, e43.

Zahirieh, A., Nesbit, M. A., Ali, A., Wang, K., He, N., Stangou, M., Bamichas, G.,Sombolos, K., Thakker, R. V. and Pei, Y. (2005). Functional analysis of a novelGATA3 mutation in a family with the hypoparathyroidism, deafness, and renaldysplasia syndrome. J. Clin. Endocrinol. Metab. 90, 2445-2450.

Zajac, J. D. and Danks, J. A. (2008). The development of the parathyroid gland: fromfish to human. Curr. Opin. Nephrol. Hypertens. 17, 353-356.

Disease Models & Mechanisms 1291

Craniofacial phenotypes in a zebrafish model of HDR RESEARCH REPORTD

iseas

e M

odel

s & M

echa

nism

s

DM

M